Cylinder Defect Identification

Key requirements are: 1. Mount multiple cameras to be able to take 360-degree pictures of the cylinder (define how many cameras are required for this purpose) 2. Perform image processing to identify whether cylinder is suitable for usage or not 3. Tag the stored result to cylinder number which is printed on cylinder


LPG cylinders move on a platform for gas filling purpose. The objective is to first identify if cylinder is suitable for use or has deformity which renders is unsuitable. Process needs to support different types of cylinders and should be able to automatically identify different types of cylinders, the entire operation needs to be done within 500 ms. This requires edge processing to be deployed and accuracy needs to be >98%



Filtering Criteria

Existing defect identification capabilities needed
The vendor should have the existing capability using image processing, with high level of accuracy. Also should be able to take up and support the implementation at different plants. Continuous learning / Machine learning is required